The general overview of read count ratio between True and False CNVs

training_rc_file   <- '/Users/rtan/terra/CNV_Espresso/result/training_set_w_rc.cnv'
training_rc_df <- read.table(file=training_rc_file,header=TRUE,sep = "\t",row.names = NULL)
head(training_rc_df)

colnames(training_rc_df)

 [1] "CHR"                     "PRED_START"              "PRED_END"                "TYPE"                    "SAMPLE"                 
 [6] "CANOES_RT"               "CLAMMS_RT"               "XHMM_RT"                 "NUM_OVERLAPS_RT"         "RD_PROP_RT"             
[11] "GC"                      "PRED_SIZE"               "MAP"                     "NUM_TARGETS"             "SIZE_LABEL"             
[16] "LABEL_VAL"               "Carriers.inGivenCohort." "RC_Ratio"               

Filter non-value rows

nrow(training_rc_df)

[1] 67499

read count number

rc_mean <- function(row){
  rc_list <- gsub("\\[||\\]|| ","",row$RC_Ratio)
  rc_value <- as.numeric(strsplit(rc_list,split=',')[[1]])
  rc_mean_value <- mean(rc_value)
  return(rc_mean_value)
}

nrow(false_cnv_df)

[1] 10580